QPSO算法在桥梁健康监测中的应用研究

发布时间：2018-04-05 15:16

本文选题：量子粒子群算法　切入点：模态参数识别　出处：《苏州科技学院》2014年硕士论文

【摘要】：量子粒子群算法(Quantum-behaved Particle Swarm Optimization, QPSO)作为一种全局收敛的人工智能算法,以其系统辨识能力强、控制参数少、善于解决复杂优化问题等优点,近年来成为智能算法应用于结构健康监测的新兴热点课题。本文依据QPSO算法自身特性,以当前模态参数和损伤识别存在的不足为出发点,将QPSO算法分别与连续小波变换(Continuous Wavelet Transform, CWT)、广义柔度矩阵(Generalized Flexibility Matrix、时程响应(Time History Response, THR)等相结合,从而将识别问题转化为优化问题,简化了识别过程,提高了识别精度。具体研究内容及其成果如下：(1)系统评述了目前结构模态参数和损伤识别的理论意义、应用背景、研究现状以及存在的问题,由此提出基于QPSO算法的桥梁健康监测研究的思路和主要内容；(2)详细阐述了QPSO算法的产生背景、基本原理、计算流程及优化特性等,同时对CWT、GFM的基本理论、主要特性和表现形式等进行了简要论述;(3)针对环境激励下小波变换和PSO算法(Particle Swarm Optimization)识别模态参数过程的优缺点,提出了QPSO+CWT的模态参数识别方法。通过对结构响应信号进行连续小波变换,实现多自由度系统响应转化为多个单自由度响应,然后QPSO算法优化单自由度响应,一次性识别出结构的模态参数。该方法一方面减少了QPSO算法所需识别的参数数量,放宽了各粒子约束条件,另一方面对模态参数共同优化,简化了计算步骤,改善了识别精度。采用不同结构类型的数值仿真对该方法有效性和可行性进行了验证；(4)基于柔度灵敏度和广义柔度灵敏度在损伤识别中存在的问题,提出了QPSO+GFM识别结构损伤的方法。利用损伤前后结构广义柔度差矩阵与结构物理参数变化之间的关系,以广义柔度矩阵构造目标函数,QPSO算法进行优化识别,达到同时准确定位、定量识别损伤的双重效果。不同结构类型的数值仿真结果表明该方法优于柔度灵敏度、广义柔度灵敏度法,其有效性不言而喻；(5)考虑到基于时程响应的PSO算法识别结构损伤的不足,提出采用直接优化损伤参数的QPSO+THR法同时识别结构损伤位置和损伤程度,通过不同结构类型数值模拟验证了该方法的有效性与良好的抗噪性。
[Abstract]:As a globally convergent artificial intelligence algorithm, Quantum-beved Particle Swarm Optimization (QPSO) has the advantages of strong system identification, less control parameters, and good at solving complex optimization problems.In recent years, intelligent algorithm has become a new hot topic in structural health monitoring.Based on the characteristics of QPSO algorithm and the shortcomings of current modal parameters and damage identification, this paper combines QPSO algorithm with continuous Wavelet transform, CWTT, generalized Flexibility matrix, time-history response time History response, etc.Thus, the identification problem is transformed into an optimization problem, which simplifies the identification process and improves the recognition accuracy.The research contents and results are as follows: (1) the theoretical significance, application background, research status and existing problems of structural modal parameters and damage identification are systematically reviewed.In this paper, the idea and main content of bridge health monitoring based on QPSO algorithm are put forward. The background, basic principle, calculation flow and optimization characteristics of QPSO algorithm are described in detail. At the same time, the basic theory of QPSO algorithm is discussed.Aiming at the advantages and disadvantages of wavelet transform and PSO algorithm Particle Swarm optimization in identifying modal parameters, a modal parameter identification method of QPSO CWT is proposed.By using continuous wavelet transform to the structural response signal, the multi-degree-of-freedom system response is transformed into multiple single-degree-of-freedom responses, and then QPSO algorithm optimizes the single-degree-of-freedom response and identifies the modal parameters of the structure at one time.On the one hand, the method reduces the number of parameters identified by the QPSO algorithm and relaxes the constraint conditions of each particle. On the other hand, it optimizes the modal parameters together, simplifies the calculation steps and improves the recognition accuracy.The effectiveness and feasibility of this method are verified by numerical simulation of different structural types. Based on the problems of flexibility sensitivity and generalized flexibility sensitivity in damage identification, a method of structural damage identification by QPSO GFM is proposed.Based on the relationship between the structural generalized flexibility matrix before and after damage and the change of structural physical parameters, the objective function QPSO algorithm is constructed with the generalized flexibility matrix for optimal identification, which can achieve the dual effects of accurate location and quantitative identification of damage at the same time.The numerical simulation results of different structural types show that this method is superior to the flexibility sensitivity, and the generalized flexibility sensitivity method, whose effectiveness is self-evident, takes into account the shortcomings of the PSO algorithm based on time-history response to identify structural damage.QPSO THR method, which directly optimizes the damage parameters, is proposed to identify the damage location and damage degree simultaneously. The effectiveness and good noise resistance of the method are verified by numerical simulation of different structural types.
【学位授予单位】：苏州科技学院
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U446

【相似文献】